Linear regression in numpy and scipy

Note that there are multiple numpy/scipy functions that do regression, fitting, etc. Of these, scipy.stats.linregress fits a line and forces an intercept. You do not have to explicitly add 1s or anything. numpy.linalg.lstsq does plain old linear regression - your inputs can even be matrices. It simply returns argmin |ax - b|^2 for given a and b, and therefore does not force an intercept. The last one I want to mention is scipy.optimize.leastsq. This one is a non-linear least squares solver, and I know nothing more about it.